2026-10-19

Trench lessons — 2026-10-19

Key takeaways
  • DexTrading, top 30 Solana wallets over 30 days (Oct 18): slow wallets are the copyable ones, but they live on one coin, and the board barely persists
  • Platform watch: nothing breaks

DexTrading, top 30 Solana wallets over 30 days (Oct 18): slow wallets are the copyable ones, but they live on one coin, and the board barely persistsevidence: C vendor, n=30

  • Numbers: median win rate 36%; for 20 of 30 the best trade was ≥50% of the month, and for 8 it was more than the whole month (everything else lost). Median best-trade share was 87.7% for slow (≥1 h hold) wallets vs 33.3% for fast ones, and 7 of the 8 "everything else lost" wallets were slow. Only 7 of 30 were also on the 7-day board, and 10 hadn't traded in 48 h. Four coins (SI, ALLINU, EMBER, e/acc) were the best trade of 14 wallets. Their gap table: at a 5% worse fill each side plus 1% fees, break-even needs +12.8% per trade; at 10%, +24.7%.

Platform watch: nothing breaks

  • Meteora: only Referral Staking cycle 4 ($10 fee minimum per LP position). Nothing for the bot.
  • Photon: performance release only; Axiom/GMGN: nothing that touches execution found.
  • No new Token-2022 fee incidents, Raydium LaunchLab changes, or Jito tip changes found for Oct 18–19.

Sources

  1. https://dextrading.com/academy/is-copy-trading-profitable
  2. https://tokentimes.in/news/pump-fun-expands-into-hyperliquid-powered-perpetual-futures-trading-1o4h95/
  3. https://github.com/pump-fun/pump-public-docs/commits
  4. https://solanahub.de/en/news/meteora-referral-staking-zyklus-4/

Evidence grades: A = large dataset, B = official docs, C = vendor/thread/small sample. Research notes for education only. Not financial advice.